一种高空俯视场景下视频图像去噪方法及装置
By simulating the visual pathway of birds, and employing a multi-channel three-dimensional spatiotemporal Gabor filter and a saliency calculation loop of the bird brain network, the problems of low efficiency in background noise suppression and small target detection in image denoising under high-altitude overhead scenes are solved, and efficient target extraction under complex backgrounds is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TSINGHUA UNIVERSITY
- Filing Date
- 2024-03-26
- Publication Date
- 2026-07-17
AI Technical Summary
In high-altitude overhead scenes, existing image denoising algorithms struggle to effectively distinguish between background noise and small targets, resulting in low target detection efficiency. In particular, they are unable to suppress noise interference and extract salient small targets in complex backgrounds.
To simulate the visual pathway of birds, a motion feature map denoising module is constructed using a multi-channel three-dimensional spatiotemporal Gabor filter and a saliency calculation loop of the bird brain network. Through latent state calculation, suppression calculation, and enhancement calculation, background noise is cyclically suppressed and saliency targets are enhanced.
It improves target detection efficiency in complex backgrounds, effectively suppresses background noise, extracts salient small targets, and has the advantages of few hyperparameters and strong generalization performance.
Smart Images
Figure CN118154454B_ABST